The proposed study aims to examine the effectiveness of four social learning treatment strategies for reducing early adolescent drug use. Essentially this research will address sets of clinical questions: A) To what extent is it feasible to eliminate or reduce early adolescent drug use within a social learning treatment framework? B) What is the optimal social learning treatment strategy for reducing early adolescent drug use? C) To what extent do studies that identify variables etiologically related to early adolescent drug use provide treatment guidelines, in terms of providing clinical targets for change? The proposed treatment study is a five year project, with the first year involving extensive pilot testing and revision of the Parent and Peer Training condition on 20 families, and the implementation and evaluation study beginning in the second year. The primary clinical sample will consist of 120 11- to 14-year-old children (grades 6 through 8) referred for substance use within the Eugene-Springfield area. Inclusion criteria include at least one documented report of the child's current drug abuse, police contact for nonstatus offenses, or severe psychopathology in the parent or child. Referred families will be randomly assigned to one of four clinical conditions.: (1) Parent Training only; (2) Peer Training only; (3) Coordinated Parent and Peer Training; and (4) Self-Directed Treatment, where parents and children are provided with the informational content of the Parent and Peer Training treatments, but do not during, at termination of treatment, six months, one year, and two years following treatment. A multiagent and multimethod approach to assessment will be used to develop multiple indicators for each theoretical construct, including constructs assumed to mediate treatment outcome (e.g., parent monitoring, peer relations, deviant peer influence, social skills) as well as the dependent variables (problem behavior, drug use). Treatment process data will be collected to further study the clinical processes most associated with beneficial outcomes within and across the treatment conditions.